Writing effective prompts
The more specific the prompt, the closer the result is to what you have in mind. This guide covers how to structure prompts for both Playground (single screen) and Studio (multi-screen app).
What to include
A good prompt covers four things:
The app type
What kind of app is this for? Fitness, finance, food delivery, social, travel, health, context shapes all the design decisions.
e.g. A fitness tracking app
What the screen or flow does
For Playground: what does this specific screen show? For Studio: what is the core user journey?
e.g. A workout detail screen showing exercises, sets/reps, and a start button
Key UI elements
List the components that matter. Don't over-specify, give the AI room to make good design decisions.
e.g. A sticky header, a list of exercise cards with checkmarks, a fixed bottom CTA
Visual tone
Dark and bold? Clean and minimal? Vibrant and playful? A reference app name works well here.
e.g. Dark, minimal, like Nike Training Club
Playground prompt example
For generating a single screen:
"A dark-themed Workout Detail screen for a fitness app. Shows the workout name at the top, a row of 3 muscle group tags, a scrollable list of 5 exercises with sets, reps, and a checkmark toggle for each, and a sticky Start Workout button at the bottom. Minimal and bold, similar to the Nike Training Club app."
Studio prompt example
For generating a full multi-screen app flow:
"A habit tracking app. Home screen shows today's habits with progress rings and a streak count. Habit detail screen has a 30-day calendar with streak highlights. Stats screen shows weekly completion charts. Add Habit screen is a simple form with name, frequency, and reminder time. Clean, minimal white design with a green accent."
Studio figures out the screen count, navigation structure, and shared design system from your description, you don't need to specify those details.
Attaching a reference image (Studio)
Studio accepts an image alongside your text prompt. Use this to give the AI a visual reference: a screenshot of an app you like, a rough wireframe, or a hand-drawn sketch.
Images work especially well for:
- Matching a specific layout or component arrangement that's hard to describe in words
- Recreating the color palette or visual style of an existing app
- Giving context from a wireframe or user flow diagram
The AI uses the image as a visual reference, not as a pixel-perfect template. Combine it with a descriptive text prompt for the best results.
Using follow-up messages
After the initial generation, describe what to change in plain language. Each follow-up is targeted, only the referenced parts are updated.
- "Make the background darker and use coral as the accent color"
- "Replace the list with a card grid, 2 columns"
- "Add a search bar at the top"
- "The header is overlapping the status bar, fix that"
- "Make the font size larger throughout"
Tips that get better results
- Name the screen type"Home screen", "Profile screen", "Checkout screen", screen names that developers and designers use give the AI useful context.
- Reference a real app"Similar to Duolingo's lesson view" or "like Airbnb's listing card", the AI understands popular app design patterns.
- Describe the content, not the implementationSay what should be shown and how it should feel. Don't try to specify layout props or exact measurements.
- One screen at a time in PlaygroundIf you want multiple screens, use Studio. Trying to describe many screens in a single Playground prompt produces worse results than generating them individually.
What RNBlocks generates for
RNBlocks is built for React Native mobile apps, iOS and Android. It generates real mobile app screens and flows, not web apps or games.
It works best for:
- Consumer and B2B mobile apps
- App ideas across any category, fitness, finance, social, productivity, e-commerce, health, travel
- Prototyping and validation before writing production code
RNBlocks does not generate web applications, game engines (Unity, Unreal, GTA-style games), or desktop software. It generates React Native apps that run on iOS and Android.